WO2001097419A1

WO2001097419A1 - Method for generation of training sequence in channel estimation

Info

Publication number: WO2001097419A1
Application number: PCT/CN2001/000860
Authority: WO
Inventors: Wei Tang; Zhaohui Cai; Zexian Li
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2000-06-07
Filing date: 2001-05-24
Publication date: 2001-12-20
Also published as: EP1300969A1; JP2004503985A; AU2001289498A1; CN1327312A; US7376115B2; US20040032849A1; EP1300969A4; CN1146156C

Abstract

The invention relates to a method for generation of training sequence in channel estimation, for dynamically determining offset amount when selecting training sequence from basic codes, according to actual operation state in specific time burst in the system, thereby generating the training sequence and allocating to each user in the specific time burst. While dynamically selecting the training sequence, the offset amount be made long as possible. The actual operation state in the system can be channel estimation state for actual user number in the specific time burst and/or each user before the specific time burst. The invention can achieve better channel estimation effect, thereby reducing code error rate, enhancing performance of reception signal in the system, and improving communicaion quality.

Description

信道估计中训练序列的生成方法 Method for generating training sequence in channel estimation

技术领域本发明涉及无线通信***中用户的训练序列的生成技术及其分配方法，具体地说涉及到无线通信***中尤其是 TD - CDMA (时分 -码分多址）或 TD - SCDMA 的训练序列的生成技术。 TECHNICAL FIELD The present invention relates to a technique for generating training sequences of users in a wireless communication system and a method for allocating the same, and in particular, to a training sequence for a TD-CDMA (Time Division-Code Division Multiple Access) or TD-SCDMA in a wireless communication system. Generation technology.

发明背景 Background of the invention

不同于单纯的 TDMA和 CDMA***， TD - CDMA***在每个时隙内包含有若干个码道，用于区分相同时隙内的不同用户。采用 TDMA与 CDMA混合的多址方式与单纯采用 TDMA或 CDMA的***相比具有性能上的优势。在 TD - CDMA***的接收机中既需要分离每帧中的时隙，也需要分离同一时隙中的不同码道（可以采用相关的处理方法）。利用这些方法，在接收端就可以将不同的信道区分开。在*** 发送端发出的信号经过传输后，空间的时变信道会对传输信号加入相应的干扰，而且移动信道的多径干扰和信道还会引起码间干扰（ IS I )和多址干扰（MAI ) 。因此为了在接收端恢复出正确的发射信号，必须正确得到信道的冲激响应，然后利用接收到的信号和所得到的信道冲激响应正确估计出所发射的信号。信道估计质量对于***性能具有举足轻重的影响。 Unlike pure TDMA and CDMA systems, TD-CDMA systems contain several code channels in each time slot, which are used to distinguish different users in the same time slot. The multiple-access method using TDMA and CDMA has a performance advantage over a system using only TDMA or CDMA. In the receiver of the TD-CDMA system, it is necessary to separate both the time slot in each frame and the different code channels in the same time slot (the relevant processing methods can be adopted). With these methods, different channels can be distinguished at the receiving end. After the signal sent by the system's transmitting end is transmitted, the spatial time-varying channel will add corresponding interference to the transmission signal, and the multi-path interference of the mobile channel and the channel will also cause inter-symbol interference (IS I) and multiple-access interference (MAI) ). Therefore, in order to recover the correct transmitted signal at the receiving end, the impulse response of the channel must be correctly obtained, and then the received signal and the obtained channel impulse response are used to correctly estimate the transmitted signal. The quality of channel estimation has a significant impact on system performance.

在现代无线通信***中，数据信号经过无线信道后，由于多径衰落和时延扩展的影响，接收到的信号发生频率选择性衰落、时间选择性衰落，信号波形发生变形，信噪比较低，难以正确恢复发射的信号。在许多***中，采用训练序列（根据位置不同，称作 preamble码或 midamb l e码）来实现信道估计，因此训练序列的生成方法的确定至关重要。 In modern wireless communication systems, after a data signal passes through a wireless channel, due to the effects of multipath fading and time delay spread, the received signal undergoes frequency selective fading, time selective fading, signal waveform distortion, and low signal-to-noise. It is difficult to correctly recover the transmitted signal. In many systems, training sequences (called preamble codes or midamblie codes according to different positions) are used to implement channel estimation, so the determination of the method of generating the training sequences is very important.

\ 在参考文献 [1]，即， Bernd Steiner and Peter Jung "Upl ink Channel Est imat ion in Sychronous CDMA Mobi le Radio Systems With Joint Detect ion" , PIMRC/ 93中，公开了一种最大似然的无偏信道估计方法和匹配滤波器的有偏的信道估计方法，同时给出了训练序列的生成方法，即认为同一个时隙中不同用户的训练序列码是基于同一个基本码或周期基本码产生的，相互间具有不同的时间移位，参考文献 [1]假定各用户的信道冲击响应具有相同的长度，则各用户的训练序列码的相对时间移位是等间隔的。 \ In reference [1], that is, Bernd Steiner and Peter Jung "Upl ink Channel Est imat ion in Sychronous CDMA Mobi le Radio Systems With Joint Detect ion", PIMRC / 93, discloses a maximum likelihood unbiased channel The estimation method and the biased channel estimation method of the matched filter, and the training sequence generation method are also given, that is, the training sequence codes of different users in the same time slot are considered to be generated based on the same basic code or periodic basic code. There are different time shifts between each other. Reference [1] assumes that the channel impulse response of each user has the same length, and the relative time shift of the training sequence code of each user is equally spaced.

图 1和图 2表示了作为现有技术的参考文献 [1] ,利用对基本码进行逐次移位获得不同用户训练序列的示意图，其中从基本码中生成训练序列时的偏移量对应于估计窗长。这种情况下不论实际工作用户数是否达到最大用户数，各用户的估计窗长都选择为即对应于同时工作用户数达到最大的情况。 Figures 1 and 2 show the reference [1] of the prior art, which is a schematic diagram of different user training sequences obtained by successively shifting the basic code. The offset when generating the training sequence from the basic code corresponds to the estimation The window is long. In this case, regardless of whether the actual number of working users reaches the maximum number of users, the estimated window length of each user is selected to correspond to the situation where the number of simultaneous working users reaches the maximum.

图 1的基本码是非周期的，其中估计窗长 w满足 W = PAR：，为最大同时工作用户数， ³为基本码长。图 2的基本码是周期的，其中估计窗长 W满足 The basic code in Figure 1 is non-periodic, where the estimated window length w satisfies W = PAR:, is the maximum number of concurrent users, and ³ is the basic code length. The basic code in Figure 2 is periodic, where the estimated window length W satisfies

K为最大同时工作用户数， P为基本码的周期。基本码选为周期性的，能进一步简化接收端进行信道估计时处理的复杂度。两种基本码的情况在原理上没有区另l，并不影响本发明所述的方法。 K is the maximum number of simultaneous working users, and P is the period of the basic code. The basic code is selected to be periodic, which can further simplify the processing complexity when the receiver performs channel estimation. The case of the two basic codes has no difference in principle, and does not affect the method described in the present invention.

以上从基本码中生成训练序列的方法，都是假定各用户的信道冲击响应具有相同的长度，并且认为各用户的训练序列码的相对时间移位是等间隔的。这种生成方法不能充分发挥信道估计的性能，因此也不可避免地影响整个通信系统的性能。 The above methods for generating a training sequence from a basic code assume that the channel impulse response of each user has the same length, and consider that the relative time shift of the training sequence code of each user is equally spaced. This generation method cannot fully exert the performance of channel estimation, and therefore inevitably affects the performance of the entire communication system.

发明内容 Summary of the Invention

本发明的提出了一种信道估计中训练序列的生成方法，根据***在特定时间突发中的实际工作状态，动态地确定从基本码或周期基本码中选取训练序列时的偏移量，由此生成训练序列并分配给该特定时间突发中的各个用户。较佳地，在所述动态选取训练序列的同时，使所述偏移量尽量长。 The present invention proposes a method for generating a training sequence in channel estimation, and dynamically determines an offset when a training sequence is selected from a basic code or a periodic basic code according to the actual working state of the system in a specific time burst. This generates a training sequence and assigns it to each user in that particular time burst. Preferably, while the training sequence is dynamically selected, the offset is made as long as possible.

较佳地，所述的***实际工作状态是指，在该特定时间突发中的实际用户数和 /或在该特定时间突发以前的各个用户的信道估计状态。 Preferably, the actual working state of the system refers to the actual number of users in the specific time burst and / or the channel estimation state of each user before the specific time burst.

本发明提出的一种根据***工作状态，动态选取训练序列的生成方法，由基站根据***中用户所处环境和信道状况，动态的选取序列并分配给各个用户，使得信道估计的窗长尽量长，达到充分利用资源，作出更青确的信道估计，同时使得***容量尽量大的目的。 The invention proposes a method for dynamically selecting training sequence generation according to the working status of the system. The base station dynamically selects sequences and assigns them to each user according to the environment and channel conditions of users in the system, so that the window length of channel estimation is as long as possible. To achieve the purpose of making full use of resources and making more accurate channel estimates, while making the system capacity as large as possible.

虽然本发明的技术方案主要针对码分多址的无线通信***，但是也同样适用于釆用类似传输结构的频分多址和时分多址的***，任何具有信号处理、通信等知识背景的工程师，都可以根据本发明设计不同用户的训练序列。 Although the technical solution of the present invention is mainly directed to a code division multiple access wireless communication system, it is also applicable to systems using frequency division multiple access and time division multiple access with similar transmission structures. Any engineer with a knowledge background in signal processing and communication, etc. , Can design training sequences for different users according to the present invention.

附图简要说明 Brief description of the drawings

为使本发明的目的、方案和优点更加清楚明白，以下举实施例，并结合附图，对本发明进一步详细说明。其中 In order to make the objectives, solutions, and advantages of the present invention clearer and clearer, the embodiments are described below in combination with the accompanying drawings to further describe the present invention in detail. among them

图 1示出了现有技术中从非周期性基本码中生成训练序列的示意图；图 2示出了现有技术中从周期性基本码中生成训练序列的示意图； FIG. 1 shows a schematic diagram of generating a training sequence from an aperiodic basic code in the prior art; FIG. 2 shows a schematic diagram of generating a training sequence from a periodic basic code in the prior art;

图 3a示出了作为本发明的一个较佳实施例的，一个时间突发在发送端的突 ' 发结构； FIG. 3a shows a burst structure of a time burst at the transmitting end as a preferred embodiment of the present invention;

图 3b示出了作为本发明的一个较佳实施例的，一个时间突发在接收端的突发结构； FIG. 3b shows a burst structure of a time burst at the receiving end as a preferred embodiment of the present invention;

图 4示出了在本发明的一个较佳实施例中，由同一周期基本码生成不同用户训练序列的方法。 FIG. 4 shows a method for generating different user training sequences from the same period basic code in a preferred embodiment of the present invention.

实施本发明的方式 Mode of Carrying Out the Invention

图 3a给出了 TD - CDMA***的一种发送的突发结构。通常情况下所说的一条链路包括发射的突发信号经过发射滤波器的处理后，经过无线信道传输到达接收机。取传播信道的固定长度为 W，每个突发结构中至少包含一个数据块和一个预先定义的训练序列块。通常情况下，利用训练序列进行的信道估计可以通过求解由已知的训练序列和接收的信号向量组成的方程组来完成。本发明的主要目的就是提出一种生成和分配训练序列的方法。在图 3a中给出了发射的 TD - CDMA的突发结构，其中包含有两个数据块和一个训练序列块，第一和第二部分数据块由中间的训练序列块隔开。假定选取固定信道的延时扩展为 T 个码片宽度，因此在接收端接收到的突发结构如图 3b中所示。其中数据块 1的部分接收信号个码片长度）移到了训练序列对应的接收信号部分，而训练序列的部分接收信号也移到了数据块 2的接收信号中。 Figure 3a shows a transmission burst structure of a TD-CDMA system. Usually said one After the burst signal transmitted by the link is processed by the transmission filter, it is transmitted to the receiver via the wireless channel. The fixed length of the propagation channel is W, and each burst structure includes at least one data block and a predefined training sequence block. Generally, channel estimation using a training sequence can be accomplished by solving a system of equations consisting of a known training sequence and a received signal vector. The main purpose of the present invention is to propose a method for generating and distributing training sequences. The burst structure of the transmitted TD-CDMA is shown in Figure 3a, which contains two data blocks and a training sequence block, the first and second part of the data block are separated by the middle training sequence block. Assume that the delay of the fixed channel is selected to be T chips wide, so the burst structure received at the receiving end is shown in Figure 3b. Among them, part of the received signal of the data block 1 (chip length) is moved to the received signal part corresponding to the training sequence, and part of the received signal of the training sequence is also moved to the received signal of the data block 2.

为了说明方便，下面以 midamble码为例说明训练序列的选取，实际上，本发明并不仅限于 midarable码的选取，其可适用于任何基于训练序列的信道估计中对训练序列的选取。并且适用于从周期性基本码或者非周期性基本码中生成训练序列。假定***中有 K个无线用户，则所讨论的***模型中即包括 K个无线信道，假定其复值的脉冲响应为： For the convenience of description, the following uses midamble codes as examples to describe the selection of training sequences. In fact, the present invention is not limited to the selection of midarable codes. It can be applied to the selection of training sequences in any channel estimation based on training sequences. It is also suitable for generating training sequences from periodic basic codes or aperiodic basic codes. Assume that there are K wireless users in the system, then the system model in question includes K wireless channels, and assume that its complex-valued impulse response is:

^ϋ , LK (1) 其长度为^。将 Κ个用户的信道响应写为向量形式： ^ ϋ, LK (1) Its length is ^. Write the channel response of K users as a vector:

h = (h^^T, h^^T...h^^T)^T (₂) 未知信道系数的总共数目为 h = (h ^ ^T , h ^ ^T ... h ^ ^T ) ^T ( ₂ ) The total number of unknown channel coefficients is

U - KW (3) 第个用户的训练序列为 U-KW (3) The training sequence for the first user is

由于假定信道冲激响应的长度为因此信道码间干扰起源于数据块中的数据符号的最后的 -1个元素，接收信号中的最初的 -1个抽样受训练序列的延时信号的影响，即存在码间干扰（ ISI ) , 不作为信道估计的一部分。因此，每个训练序列的最初的 W-1个元素的能量并不完全用于信道估计。由训练序列本身唯一确定的接收信号仅仅有个元素。设实际的接收信号为

Because the length of the channel impulse response is assumed to be, the channel-to-symbol interference originates from the last -1 element of the data symbol in the data block, and the first -1 samples in the received signal are trained The influence of the sequence's delayed signal, that is, the presence of inter-symbol interference (ISI), is not used as part of the channel estimation. Therefore, the energy of the first W-1 elements of each training sequence is not completely used for channel estimation. The received signal uniquely determined by the training sequence itself has only one element. Let the actual received signal be

根据 (4)中的训练序列 , LxW的矩阵为 According to the training sequence in (4), the matrix of LxW is

= ( ))， U (6a) = ()), U (6a)

G •ij =一 O '"M¾-j'， l..丄; .W (6b) G • ij = one O '"M¾-j', l .. 丄; .W (6b)

而 L x U 矩阵 G为 And the L x U matrix G is

其零均值的可加平稳噪声为 The addable stationary noise of its zero mean is

从而接收信号表示为 The received signal is thus expressed as

因而信道估计的问题实际上就相当于已知式（9) 中的 G、 e来求出 A的过程，其中的矩阵 G是由各个用户的训练序列组成的矩阵，向量 e是对应于训练序列部分的接收信号。 Therefore, the problem of channel estimation is actually equivalent to the process of finding A and G in eq. (9), where the matrix G is a matrix composed of the training sequences of each user, and the vector e is corresponding to the training sequence. Part of the received signal.

参考文献 [1]中为了尽量的减少***的资源的耗费，选择特殊的训练序列生成方式，即，同一个时隙中不同用户的 midamble码是基于同一个基本码或者周期基本码产生的，相互间具有不同的时间移位，令： In reference [1], in order to minimize the consumption of system resources, a special training sequence generation method is selected, that is, the midamble codes of different users in the same time slot are generated based on the same basic code or periodic basic code, and each other Have different time shifts between:

m^{m_nm₂,...,m_L^_(K__l)W)^T , m ^ (m _n m ₂ , ..., m _L ^ _(K _ _{l) W} ) ^T ,

m, £{1,-1}；£ = l,...,(L_m +(K- 1)W) 表示一个用户的基本码， m, £ {1, -1}; £ = l, ..., (L _m + (K-1) W) represents the basic code of a user,

其中， m_; = m^ = ( + 1)"..，(L„, +{K- 1)W) , 是 midamble码的长度，为基本码的周期，表示信道估计窗（即信道脉冲响应长度）的大小。那么个不同用户的 midamble的产生基于公式： m = m_i+(_K-k)W, i = ' ;L_m; k = l, - - -,K₀ Where m _; = m ^ = (+ 1) ".., (L„, + {K-1) W), is the length of the midamble code, is the period of the basic code, and represents the channel estimation window (that is, the channel impulse response Length). The generation of midamble for different users is based on the formula: m = m _{i +} ( _K -k) W, i = '; L _m ; k = l,---, K ₀

这样（7 )式中的矩阵 G的各列是由一个训练序列 /η,·的不同平移得到，即矩阵 G 是循环矩阵。 In this way, the columns of the matrix G in (7) are obtained by different translations of a training sequence / η, ·, that is, the matrix G is a cyclic matrix.

这样构成的 ffiidamble码可以通过做一次相关运算实现多个用户的信道估计。其中，选取用户的 midamble时的偏移量 W值，是一个比较重要的参数，而且在确定的物理层工作过程下，该值应该是确的。实际中 W值的选取需要考虑下面几个因素：帧格式中 midamble码的长； tL_w；选取的周期基本码的周期 i⁵ 的大小；一个时隙中同时工作的最大用户数移动信道的延迟扩展量；移位前同时使用 ra i darab 1 e码的数量。如杲假设同一个时隙内同时工作的最大用户数，信道冲激响应长度的选取过程应为： The ffiidamble code formed in this way can achieve channel estimation for multiple users by performing a correlation operation. Among them, the value of the offset W when the user's midamble is selected is a relatively important parameter, and the value should be correct under the determined physical layer working process. The selection of the W value in practice needs to consider the following factors: the length of the midamble code in the frame format; tL _w ; the size of the period i ⁵ of the selected periodic basic code; the maximum number of users working simultaneously in one slot; the delay of the mobile channel Spread amount; the number of ra darab 1 e codes used simultaneously before shifting. For example, assuming the maximum number of users working simultaneously in the same time slot, the selection process of the channel impulse response length should be:

1. 为了快速实现信道估计，根据最大似然的信道估计方法（参考文献 [1] ) , 应使构造的估计矩阵是循环矩阵，所以应首先使等式 w = 」d成立，其中 " L 」，，表示向下取整数操作。 1. In order to achieve channel estimation quickly, according to the maximum likelihood channel estimation method (reference [1]), the constructed estimation matrix should be a cyclic matrix, so the equation w = "d" should be established first, where "L" ,, Represents the integer operation down.

2. 估计矩阵的列数即接收端收到的用于信道估计的 chip数）和 midamble 码的长关系应该满足 L + 1 = 。 2. The number of columns of the estimation matrix (that is, the number of chips received by the receiver for channel estimation) and the length relationship between the midamble code should satisfy L + 1 =.

3. 为了使估计有解，应该满足 ^≥ 。 3. In order for the estimation to have a solution, ^ ≥ should be satisfied.

4. 确定是否满足应用要求。 4. Determine if the application requirements are met.

其中，用户数决定了 ^能取的长度值。'一方面，根据 ITU- R M. 1225 所给出的几种典型的移动通信***的信道模型，在实际***工作过程中，信道的延返 ·Τ展量可能 ^大，另一方面由于信道是时变的，并且考虑到传输效率及保证信息速率等方面因素，数据帧格式中 midamb ie码的长度是有限的，这样在同时工作的最大用户数大于一定数目后， W的取值就会小于实际信道的延迟扩展量 W , 若某个用户信道的! T >W, 那么一方面由于丢失部分信息使得估计值不准，另一方面对取相邻移位 midamble码的用户的信道估计造成干扰。因此，茌逸取用尸的 midamb l e码的偏移量的过程中， W的取值应该是自适应变化的，根据用户数的不同和 /或具体的信道条件，相应地调整各用户的取值以及相对位置,. 使得尽可能多的用户的信道估计窗的长度（即各偏移量值）保持最长。 ' 因此，各用户在选取 midarab l e码时， W值和 raidamble码偏移序号的选取策略可以 4既括为： Among them, the number of users determines the length value that ^ can take. 'On the one hand, according to the channel models of several typical mobile communication systems given by ITU-R M. 1225, in the actual system operation process, the channel delay and T spread may be large. On the other hand, due to the channel It is time-varying, and taking into account factors such as transmission efficiency and guaranteed information rate, the length of the midambie code in the data frame format is limited. After the maximum number of users working at the same time is greater than a certain number, the value of W will be The delay spread W is less than the actual channel. If T of a user's channel is greater than T, the estimated value is inaccurate due to the loss of some information. On the other hand, the channel estimation of the user who takes the adjacent shifted midamble code causes interference. Therefore, in the process of taking the offset of the midamb le code of the corpse, the value of W should be adaptively changed. According to the number of users and / or specific channel conditions, the user's access to the corresponding adjustment should be adjusted accordingly. Value and relative position, so that the length of the channel estimation window (i.e., each offset value) of as many users as possible remains the longest. 'Therefore, when each user selects the midarab le code, the selection strategy of the W value and the raidamble code offset number can be included as:

在已有的用户数目一定的情况下使其 midamb le码的距离尽量远；并且对新增加的用户，根据已有用户所选的训练序列和已有用户各信道的延时扩展量，选最合适的 midamble码偏移序号。下面，以一个时隙中有一个基本码为例，具体说明作为较佳实施例的三种训练序列的生成方法，其中 K表示最大工作用户数， it表示具体偏移序号，选取的周期基本码的周期为 P， midamb l e码的长度： In the case of a certain number of existing users, make the distance of the midamb le code as far as possible; and for newly added users, according to the training sequence selected by the existing user and the delay spread of each channel of the existing user, choose the most A suitable midamble code offset number. In the following, a basic code in one time slot is taken as an example to specifically describe three methods for generating training sequences as a preferred embodiment, where K represents a maximum number of working users, it represents a specific offset sequence number, and a selected periodic basic code. The period is P, the length of the midamb le code:

( 1 )当同时工作用户数为 M< ^时，初始时，的长度取为 If Ad (1) When the number of simultaneous working users is M <^, initially, the length is taken as If Ad

M。工作过程中，网络端根据各用户信道延迟扩展量的不同，可以重新调整偏移量 k=l , . . . , Μ, 使得信道延迟扩展量小的用户占用小的偏移量，使得信道延迟扩展量大的用户占用大的偏移量。如果有用户退出，则随后新增加' 的用户取代原用户的位置，这是完全自适应的情况。 M. In the working process, the network end can readjust the offset k = l, ..,, M according to the difference of the channel delay spread of each user, so that users with small channel delay spread occupy a small offset and make the channel delay Large-scale users occupy large offsets. If a user quits, then a new user is added to replace the original user's position, which is a completely adaptive situation.

( 2 )当同时工作用户数为 Μ Κ时，各用户的的长度是相同的，

,(2) When the number of simultaneous working users is KM, the length of each user is the same,

,

M < K, 随着用户数的变化，取值相应变化。每个时隙中所配置的用户数可以通过***中的公共信道广播或采用其他方式通知各用户。 M <K, as the number of users changes, the value changes accordingly. The number of users configured in each time slot can be broadcasted through a common channel in the system or notified to other users by other means.

( 3 )为了实现的方便，可以设置若干个等级，每种等级对应一种用户数情况，相应分配固定的估计窗长度。例如对于以 K/2个用户为界的两个等级，当用户数大于时， W值取为 L / d ，当用户数小于时， W值取为 |_2_P/ d 。图 4给出最大用户数为 16，但是当前实际用户数为 4情况下的一种自适应训练序列的生成方法，其中基本码表示确定的序列，可以釆用伪随机序列或者通过计算机按一定要求搜索产生。这时各用户的信道估计窗的长度自适应地选为 4W，而木是在 16个用户情况下的 W, 估计窗长度的加大，对应于信道估计精确性的提高。根据上面的介绍，我们可以确定当训练序列的长度固定后，就可以根据时隙中同时工作的用户数来确定信道冲激响应的长度，同时得到各个用户相应的训练序列。如：如杲***中每个突发中训练序列的长度为 144个数据比特或码片 (扩频***中），其中周期基本码的长度为 128个数据比特或码片，同时如果每个时隙中可同时工作的最大用户数为 16，即最多可以有 16个用户同时工作在相同的时隙中，那么才艮据 =Ρ， W的值为 8 , 即用户的训练序列和用户 - 1的训练序列之间的移位为 8。当然如果每个时隙中同时工作的用户数减半，则可以根据 ^=尸的关系，不同用户之间训练序列的移位可以加倍。如此类推，用户可以在性能和复杂度之间做一个折衷，真正应用到实际***中。 (3) For the convenience of implementation, several levels may be set, each level corresponds to a situation of the number of users, and a fixed estimated window length is allocated accordingly. For example, for two levels bounded by K / 2 users, when the number of users is greater, the W value is taken as L / d, and when the number of users is less, the W value is taken as | _2_P / d. Figure 4 shows a method for generating an adaptive training sequence when the maximum number of users is 16, but the current actual number of users is 4, where the basic code represents a determined sequence, and a pseudo-random sequence can be used or a computer can meet certain requirements. A search is generated. At this time, the length of the channel estimation window of each user is adaptively selected as 4W, and W in the case of 16 users. The increase of the estimation window length corresponds to the improvement of the accuracy of the channel estimation. According to the introduction above, we can determine that when the length of the training sequence is fixed, the length of the channel impulse response can be determined according to the number of users working simultaneously in the time slot, and the corresponding training sequence of each user can be obtained at the same time. For example: The length of the training sequence in each burst in the Rugao system is 144 data bits or chips (in a spread-spectrum system), where the length of the periodic basic code is 128 data bits or chips, and if each time The maximum number of users that can work simultaneously in the slot is 16, that is, a maximum of 16 users can work in the same time slot at the same time. Then, according to P = W, the value of W is 8, which is the user's training sequence and user-1 The shift between training sequences is 8. Of course, if the number of users working simultaneously in each time slot is halved, the shift of the training sequence between different users can be doubled according to the relationship of ^ = corpse. By analogy, users can make a compromise between performance and complexity, and really apply it to actual systems.

在实际***中，也可以采用不同的训练序列的生成策略，也就是说在基站生成训练序列时满足：（1)在已有的用户数目一定的情况下使其 midamble码的距离尽量远。（2) 对新增加的用户，根据已有用户的延时。选最合适的 midamble 码值赋予。例如：根据用户数的增加小于 8时，新增加的用户，依次取， 1 , 3 , 5 , 7 , 9 , 11 , 13 , 15号 midamb le码；其中若有用户退出，则新增加的用户取原退出的用户的 midamble码；当用户超过 8时，新增加的用户取的 midarable码的序号取在巳有用户中实际信道冲激响应长度最短的用户的序号 + 1的 m i damb 1 e码 ( 设序号的增加跟移位的方向一致，实际的信道冲激响应长度可以由信道估计值直接得到），直到 16个用户配置完为止。另外一种方式是开始随用户增加的用户取值依次为： 1 , 9 , 5 , 13, 3 , 7 , 11 , 15等，这些都是训练序列生成方式的例子。该方法给出了适用于无线通信***中的训练序列的生成方法，尤其是适用于目前第三代移动通信***中的 WCDMA - TDD (宽带码分多址-时分双工）、 TD - SCDMA (时分双工 -同步码分多址）、和 TD - CDMA (时分双工 -码分多址） ***中的一种不同用户的训练序列的生成方法及选取方法。采用本发明的方法后，可以做到最大限度地利用可利用的资源，使得***的性能和复杂度达到最佳的折衷方式。 In an actual system, different training sequence generation strategies can also be adopted, that is, when the training sequence is generated by the base station, (1) the distance of the midamble code is made as far as possible when the number of existing users is constant. (2) For newly added users, according to the delay of existing users. Choose the most appropriate midamble code value to assign. For example: When the increase in the number of users is less than 8, the newly added users are taken in turn, and the codes are 1, 3, 5, 7, 9, 9, 13, 15, 15; if any users log out, the newly added users Take the midamble code of the user who originally logged out; when the user exceeds 8, the serial number of the midarable code taken by the newly added user is the serial number of the user with the shortest actual channel impulse response length among the users + the mi damb 1 e code (It is assumed that the increase of the serial number is consistent with the direction of the shift, and the actual channel impulse response length can be directly obtained from the channel estimation value), until 16 users are configured. Another way is to start adding user values as the users increase in order: 1, 9, 5, 13, 3, 7, 11, 15, etc. These are all examples of training sequence generation methods. This method gives a training sequence generation method suitable for wireless communication systems, especially for WCDMA-TDD (Wideband Code Division Multiple Access-Time Division Duplex), TD-SCDMA ( A method for generating and selecting a training sequence for different users in time division duplex-synchronous code division multiple access) and TD-CDMA (time division duplex-code division multiple access) systems. After the method of the present invention is adopted, the available resources can be used to the maximum, so that the performance and complexity of the system reach the best compromise.

虽然本发明的技术方案主要针对码分多址的无线通信***，但是也同样适用于采用类似传输结构的通信***，任何具有信号处理、通信等知识背景的工程师，都可以才艮据本发明设计不同用户的训练序列。 Although the technical solution of the present invention is mainly directed to a code division multiple access wireless communication system, it is also applicable to a communication system using a similar transmission structure. Any engineer with a knowledge background in signal processing and communication can design according to the present invention. Training sequences for different users.

本发明适用于通信环境较恶劣的移动通信***中，如车载移动台环境下，使用于第三代移动通信***中，可以减小误码率，提高***接收信号的性能，从而改善通信质量。 The present invention is applicable to a mobile communication system with a poor communication environment, such as an on-board mobile station environment, and is used in a third-generation mobile communication system, which can reduce the bit error rate and improve the performance of the system's signal reception, thereby improving the communication quality.

本发明并不仅限于 midamb le码的选取，其可适用于任何基于训练序列的信道估计中对训练序列的选取，包括 preamb le码的选取。并且适用于周期性基本码或者非周期性基本码。以上所述仅为本发明的较佳实施例而已，并不用以限制本发明。 The present invention is not limited to the selection of midamb le codes, and it is applicable to the selection of training sequences in any channel estimation based on training sequences, including the selection of preamb le codes. It also applies to periodic basic codes or aperiodic basic codes. The above are only preferred embodiments of the present invention, and are not intended to limit the present invention.

Claims

权利要求书 Claim

1、信道估计中训练序列的生成方法，其特征在于： 1. A method for generating a training sequence in channel estimation, which is characterized by:

根据***在特定时间突发中的实际工作状态 , 动态地确定从基本码中选取训练序列时的偏移量，由此生成训练序列并分配给该特定时间突发中的各个用户，或者直接该偏移量。 According to the actual working state of the system in a specific time burst, dynamically determine the offset when the training sequence is selected from the basic code, thereby generating a training sequence and assigning it to each user in the specific time burst, or directly Offset.

2、如权利要求 1所述的方法，其特征在于在所述动态选取训练序列的同时，使所述偏移量尽量长。 2. The method according to claim 1, wherein the offset is made as long as possible while the training sequence is dynamically selected.

3、如权利要求 1所述的方法，其特征在于所述的***实际工作状态是指，在该特定时间突发中的实际用户数和 /或在该特定时间突发以前的各个用户的信道估计状态。 3. The method according to claim 1, characterized in that the actual working state of the system refers to the number of actual users in the specific time burst and / or the channels of individual users before the specific time burst Estimated status.

4、如权利要求 1所述的方法，其特征在于所述的基本码是非周期基本码，或者周期基本码。 4. The method according to claim 1, wherein the basic code is an aperiodic basic code or a periodic basic code.

5、如权利要求 1所迷的方法，其特征在于当同时工作用户 1 最大同时工作用户数 K时，各用户偏移量取为 LP/M丄并且随着用户数的变化，取值相应变化。 . 5. The method as claimed in claim 1, characterized in that when the maximum number of simultaneous working users K is 1, the offset of each user is taken as LP / M 丄 and the value changes correspondingly as the number of users changes. . .

6、如权利要求 1所述的方法，其特征在于，设置若干个用户数等级，每种等级对应一种用户数情况，相应分配固定的偏移量，对应用户数少的等级，其相应分配的偏移量较长。 6. The method according to claim 1, wherein a plurality of levels of user numbers are set, each level corresponds to a situation of the number of users, a fixed offset is allocated correspondingly, and a level corresponding to a small number of users is allocated correspondingly. The offset is longer.

7、如权利要求 1、 2、 3、 4、 5、或 6所述的方法，其特征在于，工作过程中，网络端根据各用户信道延迟扩展量的不同，重新调整各个用户的偏移量，使得信道延迟扩展量小的用户占用小的偏移量，使得信道延迟扩展量大的用户占用大的偏移量；在有用户退出的情况下，随后新增加的用户取代原用户的位置。 7. The method according to claim 1, 2, 3, 4, 5, or 6, characterized in that, in the working process, the network end readjusts the offset of each user according to the difference in the delay extension of the channel of each user. A user with a small channel delay extension occupies a small offset, and a user with a large channel delay extension occupies a large offset; in the case of a user exiting, a newly added user subsequently replaces the original user's position.